 Hello, this is Harith, and welcome to the Seagaf Deliria meeting. We don't have a lot of agenda for today's meeting. I'm still waiting for more people to join, and I'll always say it's also over there. So if anyone want to introduce yourself to the Seag, and it's really welcome. I'll go ahead and say hello because there's other people I'm here that I don't know. I have been here before, but it's been sometime. Cornelia Davis, I'm the CTO at WeWorks. Oh, congratulations for a new move, by the way. Thank you. It's good to be here. Okay, so since we are now familiar with each other, and so let me give a quick update on today's agenda, because we're talking about some updates regarding to the some working items in the Seagaf Deliria. So the first update is regarding to, oh, let me share the screen. So we do have the first update, which is regarding to the Seagaf Deliria landscape. Oh, I will call that the Seagaf Deliria landscape. So I will add a little bit of background here, because if you look at today's Seagaf landscape, you will notice that there is a section which name is, which name is application, definition or management or something, which is really confusing, because it included, for example, database and storage as part of the application. And it's also had a very high-level category or classified regarding to the components in the application delivery ecosystem. So we began to figure out that we need to find a way to refactor this part of the landscape to the high-level idea, which we also proposed to GOC and I got feedback, is to create a new tab. So it will create, we will try to create a new tab on the landscape, which name is like application delivery or something like that. It's quite similar to what serverless have today. So it will be an entire new tab, including the several categories listed in the Dictionary of Seagaf Delivery. And so this has been proposed to GOC and we think we're all going to choose this way. But this does not stop a question, which also, I think I've raised a lot of discussion recently, is about how can we find a way to explain what's inside this landscape. For example, there's something, there's a project in the landscape, but how can we expand to any users what is projected used for and when should you use that. So this is right now for, for now, is out of scope of the Seagaf Delivery landscape. But we also think it's it's a necessary thing to do for our next working item. I think also I also think that Alois have some feedback on that. So we will, we will probably, probably discuss regarding to that part in next meeting. I don't know if Alois will join today. So we will discuss about that. So the eventually, I think the goal is that we want to publish a white paper. The white paper is sponsored by Seagaf Delivery and since we will try to expand what is happening inside this landscape, what is exactly a project is focusing on and how can users benefit from that. So this is a quick update from the Seagaf application Delivery landscape. I think the next action is we're all trying to draft a proposal for the, for the landscape ETHOP repo to highlight a a simple version of how the new tab will maybe look like. That will be the next action item. So this is the first update and the second update is regarding to the working groups and I just also talked with Alois regarding to that. So we notice that the working group, the working groups, I mean we now have actually two working groups. The first one is all further working group. The second one is air gap, air gap working group. So we notice that the two working groups didn't make a lot of progress in the past few months and I think we also think that there are some blockers to continue the existing discussions in the working groups. Some of our, some are related to the operator technician and some are related to the best practice, best practices of air gap environment application Delivery. So we are proposing that we move those discussions back to the stake. This is a general idea and not determined yet. So we just think about that. So we also would like to see what do you think if we move the discussions in the working groups back to the stake so we can focus or we can gather people discussing on those discussions instead of splitting them into different meetings. So what do we think? Any ideas regarding to that? I mean is the lack of progress, what is that telling us? I mean the whole reason that we spun off into those working groups was because it seemed like it was pretty specialized and so did we disprove that hypothesis that there was enough interest to do that? Is it lack of interest? I mean I suppose that's some of the analysis that we need to do. Why are those things stalled? Yeah I think one observation I had have recently is that so the reason we actually move those discussions to a separate working group is actually because those discussions actually raise a lot of you know I will say conflicts in the stake meetings. So a lot of things cannot be fixed, a lot of things cannot be determined. So every time we have a discussion for example related to the operator dedication it may be lost like one hour. That is the usual we had before so those discussions really need to find a place to be tangled with the expert from the operator ecosystem to fix that. But that's why we actually created an operator working group to discuss that topic. But the recent progress is which is unexpected because I think the community began to think about okay this may be an issue that cannot be solved by simply discussion or published in the white paper regarding to that because there are different interested parties for example Ray had there are people from chorus before and there are people from Ray had there are people from Microsoft and they debate with each other but it seems very hard to reach the conclusion from my perspective. I don't know how we can fix that but I think the first step we want to do is move the discussion back seek so we can see what's happening there. Maybe we want to stop some of the discussions or just or just you know try to publish something that to explain what's happening in the related to discussion. If you check the pre-risk discussion it's pretty long in the seek after every mailing list but I don't think there's conclusion regarding to that part. So I mean from a personal perspective I love the idea of moving it back into this thing because as you know it's difficult for me to make even one meeting every two weeks. It's even more difficult to do three meetings every two weeks and so I'd love to have those topics discussed in one forum. It'll be interesting to see if we end up back in the in the pickle that we were in which caused the spin-off which is now it consumes the entire agenda and it's contentious but you know it's it's hard to argue with the the logic here as well. I see yeah this is also what we also think about that maybe on to first person people on the same meeting I mean on a single meeting instead of space discussions in multiple meetings it's really hard especially considering a lot of people are working from home they have a lot of meetings every day so it will either handle the discussions regarding to that point. Okay thank you for the input. So another discussion is regarding to the air-gapped application delivery working group and this is actually a very interesting topic we mentioned that several times before and we also think it's a very useful tool or very useful technology regarding to the application delivery ecosystem but that in that part in that working group my observation is that it's still not quite there so other people have their own way to do air-gapped application delivery but there's seen a thing which we think it's still missing today that how can we do that in a common way or is there is there already technology to support that or we want to or we have to propose something to solve that problem so I think the issue is still very early stage and people have different ways to do their own air-gapped delivery in different environments so I don't know if you guys have you folks have any experience regarding to air-gapped application delivery and whether we want to how can we move this direction forward especially making become a mainstream discussion or mainstream topic in this ecosystem. Yeah I think what the group was working on is collecting a number of best practices how people are doing it I think they were not like super successful doing this I think there's one example in there still but having more examples of people who are actually doing this and engaging with them is would definitely be helpful there I'm just not sure how we get people to to really do this this is also maybe something with the end user community to engage who wants to share best practices what they're doing I think it's less about I think we're not at this stage where we write that one white paper this is how you do air-gapped deliveries I think we are more at okay this is what we did this is what worked this is what did not work this is what we're doing but with what we don't like doing on a more anecdotal point of view but unfortunately historically yeah I think we we also didn't have a lot of uptake for for these meetings there were a couple of project presentations but I think that's what we're kind of like struggling a bit with now that we don't have in general the project review so much to like getting this disengagement model model going but I think it would still be worth it has been on the agenda for a while how maybe people in the telco seek work with this because initially there's wars coming from the telcos but that would be good just collecting trying to get that number of examples up that we have available it's just funny because the reason why this working group exists was last year's north american pukecon that was the primary feedback that that harry and I got can you work on air-gapped so we started to work on air-gapped and it seems to be I think below that critical mass momentum that that that we have right now I think active outreach is probably the best that that that we can do here and telco might be our first stop and I can take this this as an action item reaching out in other words you think Cornelia what we could do yeah no I agree and I'll tell you from the weaver's perspective as I'm sure you can imagine we think that get off plays a role in air-gapped and we actually have customers folks that we're working with some some whom are quite public uh dutch telecom for example speaks publicly about the work that they're doing with flux and those types of things and we have a great relationship with some of those telcos and so I'm sitting here listening and and you weren't on earlier when I kind of semi you know had an awkward apology for not showing up to these things for quite some time I'm renewing my my lock on my calendar for these meetings um I would love to work on this um and put some cycles into it because we we have customers that would be very keen um so that we can represent so yeah I think we should work on this and I think you're right I think telcos is a really great place to start yeah so then as an action if you could maybe get some of your customers interested to to present and maybe share that success story we have a document in case you I'll put it in the meeting notes later or you can reach out to me directly if you can find it put the working group hat in there and I'll reach out directly also to the telco sick regarding this so maybe we can I think for us it's like really building up momentum right now like we are in a like weird time right now where as I said we mentioned a lot of people and a lot of meetings so value creation is more important than ever to get people into yet another zoom call and also making content consumable in a different format is I think helpful so okay I'm taking giving myself an action item I think that's also a lesson learned from me to some extent in the past we we're just talking about stuff but not action iteming yeah I'm just tapping up the notes here sorry I missed your name here luckily Google is smart um yeah and yeah I think that's good just let's try to move it there yep I mean one thing that could also easily be done is just posting it on the TSE mailing list at some point if you really want to work on and we have some use cases and we want to do more it might just turn out to be a bit of spam but it's currently not super active many people are really working on those topics we might also get some feedback there so I noted a lot of people have subscribed to the sick sorry to the TSE mailing list it's not what it's actually intended for but there's not that much going on right now and we might also might get more visibility for this but I think having a couple of presentations already available is good because once you reach out to people and say hey we will talk about this it's good to say well there's already three examples in there what things look like it's not just like inviting them over to the desert of documents and let me find the document right now yeah all right I just also mentioned that they are a pretty working group and I don't know why so you have any feedback or ideas regarding to that part um so if you look at the operator working group like the operator talks so this whole thing started this is the CNCF one thing as a new operator definition that goes a bit further than the one that's currently in the Kubernetes documentation which means which model states it's a crd and the controller behind it to be more specific what to do there we have this document and I think it just needs at least that final push to get it to a v0.5 release I think it has just been sitting around this is again to some extent more in the chores area just going over this document and saying okay this is what we came up with so that we feel comfortable sharing this on the tuc in an upcoming tuc update and also back to the tuc see what they think about it uh the the problem I kind of had why this is wasn't moving forward because we always go back to like this fundamental discussions and they go very far from we say well we don't want we want to include use cases we want to include this tool and then it goes all the way we had discussions that could go went all the way up um to well I want to talk to my non-technical cto and explain to them what an operator is because I have to write it and he kind of tells me I don't have to do this because kubernetes is doing everything anyway so I think we can solve all problems for for for everybody honestly I'd love to but I think we can't so first of all I think the document number one needs cleanup and I think I'll just get a good glass of wine and just do the cleanup after this document this is just accepting exactly accepting edits and then at the very beginning I think we all should have this discussion we can put it on the next agenda item what do we really want to achieve and my intention in the beginning was somebody is writing an application on kubernetes they're here they heard about operators and they want to understand what they should be using them for or whether they should not be using them for but it like became this thing that was like this constantly moving target and at well I'd like to make the discussion but the most valuable thing for people uh to to know when it comes to operator what are the questions that people usually come up with and what they they don't understand I'm trying to find this document I haven't I also have to admit I haven't opened it in a while honestly but number one when I find it I'll really look into just getting it to the point where yeah where we have it that we can review it and discuss it the next time and I think maybe that's the best approach I'll try I'll work on cleaning up the document share it on the mailing list the next meeting we get to a conclusion what we want to have in there and what we don't want to have in there but like this really felt something that kept going in circles and whenever we had some we'd written something down somebody in the group came up with no we can't do it that way we have to go in an entirely different direction and that will never get us to a point where we read your document uh that that's that that's of value to people out there I mean no no what you think but what the value of this document would be at some point it could be a guidance and how to implement certain things because I think certain things are highly unclear how to do it ideally with an operator there are a number of best practices that could be condensed into like Microsoft has a good one how it would split out things into different operators yeah I think I've been going a bit back and forth here but number one cleaning it up and like being really clear on the purpose of this document and uh then then getting the work done I think that's why it kind of got stuck because it was at some point frustrating writing something somebody said no we have to do something completely different and obviously me not even being able to find it to right now yeah and I just dropped a link in the chat but it's not that that was just a couple of user stories um it isn't the doc that you're referring to I'm going through the I'm going through the meeting notes trying to find it anyway it doesn't seem to be linked from the notes oh mailing list yeah I can also take this as some homework I think document should be easier to find if we want to be able to engage right yeah hello it's a positive on the trade wasn't this document that's the document from um create I think that's the create document yeah that's the user story from create let's just yeah I think we can all agree that that we should have like this one page that we go to where we find all those things and it kind of doesn't exist and that should be fixed what we did in the but I haven't been in a while as well I don't know how many people know this but we actually have a github repo like a word actually all of the things could be linked to I'm just posting it in here and I think that's the and obviously we have you're kept in there that's where we would find it but we don't have ah we don't have the the operate the pieces in it that's where you found the document hopefully that's where your stories yeah lose the stories yeah but I did not find I think that's also the only one that exists honestly it's the only document mm-hmm but that's the charter and this is one yeah so yeah I found it did you just find it too I remember I think I'm checking out this project right now just creating and reading empty and I'm just posting it in there that's what that last one is the document that you have in your head operator definition working document mm-hmm so what I just did in the repo I just created and documents md file I just created a document file which I should have that and I was going to paste things in there yeah I think we have to do some thanks for helping we have to do some housekeeping here that should then also help people to to engage yeah that's the working document and it has a lot of comments and I think it just needs some it just needs some some cleanup and yeah we have a director of non-couple need this company to understand what an operator is is this really our goal I mean I think that that's a lot of unclosed comments means that we haven't reached consensus typically yeah so what I will do I'll go over the documents and see like some of them you could easily accept there's just some text formatting stuff in there and I'll at least comment to all the people also some people never followed up like now for example as Matt was complaining about the helm description and then asked him okay please can you just make it make it to what you think it should be and then people are also not replying back so I'll go over it and I'll do some housekeeping there as well I think whenever I type a link I never know the 100 correct syntax in in markup I feel like I always get the braces wrong so I'll work over this and try to create also some engagement and submit and post it to the mailing list to also get it beyond the document so this is where I started to put some some docs and obviously some more cleanup work to be done here so I actually personally feel it's very hard to reach any conclusion on the operator definition documentation because the different interest departments actually have different opinions on that so it's not hard to reach a accurate conclusion regarding to that definition and that's my personal opinion maybe that's also our feedback at some point that we tried it and we can't reach a conclusion yeah we can also write down that the conclusion is that there's no conclusion because everybody have deep opinion regarding to that that actually might be a fair conclusion at some point yeah it's true I think there are multiple issues there so the operator itself should not go into the way that compete with something like helm or package management but somehow the way had go to that direction that actually already create confusing in the technical part and because of that problem it's a decision that operator folks made and on the other hand helm should definitely focus on package management instead of trying to you know fix the day-to-operations issues so they too are are overlapping today that's why if you want to make clear separation between those two by something like operator definition I don't think it actually works there's a related topic which is actually gtops I actually think about gtops for several times so I do see there are a lot of possibilities to improve the interoperability of the gtops tools I noticed that Flux CD project is trying to working something only Flux v2 which defines several I think I will call them building blocks for example how to interact with it how to how to interact with helm how to leverage customize how to drive the state machine of the gtops workflow so they basically splits gtops into multiple building blocks and similar effort also happens to the Argo CD part and they already split some signature name is gtops engine and I think they're also discussing about how to defy I will I will call them standard interfaces between gtops and the reconciling control loop in a server side how to defy a a bunch of standard interface between the different renders like helm, customize, queue and how to add your own plugins to do that so I don't know if there are any possibilities that it can involve the effort in a seek application delivery and is anything we can help or especially when vworks folks are actually in the meeting so I also hope to see if any ideas regarding gtops work yeah no I think that's it I think that's a great idea what's interesting is of course we you know air gap we're interested in that operators are an important element of doing gtops the right way and so they're related to all the things but I think what you're bringing up is the possibility that gtops be kind of a first-class entity in that we discuss as opposed to a second-class as a part of air gap and as a part of operator working group and and all of those things and I think that I think the time might be right to do that. Yeah yeah one one trend I I actually I'm thinking that essentially that gtops plus with a plus with the CI pipeline if you combine them together you are actually I mean the whole ecosystem actually creating a fresh new CI CD workflow around the cloneative technology stack it can be compared to something like spinnaker but it's fully based on cloneative technology and it's separate CI and CD and you beat as a source of truth at the center and it's encouraged interact it encourages interoperability instead of a monolithic stank so I think that is the right direction to go it's something that CGAP delivery want to endorse and to advocate I mean but how can we do that how can we make the CIG I mean make the community engage more in this I mean gtops projects and these ideas it's something that we want to think about and I know that we works are doing very well here but how can we do that more I mean more collaboratively with the community I think that is also one direction we're thinking about it's really important to me yeah likewise yeah this is ideally I have just to say this is some ideas I have how to make the CIG collaborate more in the whole community there are some topics I think indeed that needs the involvement from the community as well as the community also to use more so this is something that we want to looking at want to look at yeah I mean what what we could what we had in the charter at some point is I think what also goes into like this whole discussion right now where everybody says the CNCF landscape is way too complex there's like a million tools I don't even know what to do with it having this idea of a test I had this idea back that we have some kind of like a test bed where we would use technologies and we could plug them in for different things and this could go together nicely like we could have one where we say okay we want to combine for example OAM with gtops and would this even fit together like we start with very atomic building blocks and then we're trying to build also the reference model that you defined here with here we start to define more like complete end-to-end more or less examples and I think people will be contributing if this is something where we have like these test beds on the the after delivery or reference whatever where the CIG and most of them have tutorials anyways so this is nothing where people have to start from from scratch like for example the the flux tutorial is it's pretty much uh available um out of the box like how to getting started with specific things and then trying to combine things together because like to that point combining OAM with a gtops type of approach is kind of interesting as well would you agree yeah I agree so so what what is this well looks like it looks like a white paper or some so what is the outcome you know like I'd like the outcome ideally to be code be code you mean samples and demos yeah so we pick a there are obviously a number of very famous sample applications that even with we start with a hello world one or at least two well and okay let's try to use this with this approach let's try to use it with this approach and that has been interesting here and we really say to your starting point what you always get is say two or three images on docker hub and the basic kubernetes primitives to deploy this application and now show how your project adds value beyond that because if it doesn't add value beyond what I already have and I just keep control apply stuff why would I be using it as an end user I see I see a point here yeah this is really interesting because it's actually a good place for application delivery stick to host a I mean something like a sample or demo ripple to show the user cases for how people use different technologies and how also best practice to use certain technology I think it's also much better than something like a white paper because I don't even pay a lot of time to read those white papers but I do usually you want to run maybe the demos yeah that is really interesting idea and also back when we discussed what we want to do a lot of things are not part of standard demo applications which you want to be part like secrets management or current credentials management like everybody's using this the stock shop or hipster example which is greatest a bit too big as our experiences we we only use a subset we only usually use the card service because we do multi-stage deployments and then it gets like kind of expensive to run it in three or four stages for a demo but certain aspects of what people have to ship are not really in there or also like the multi-stage one for me is is one that we are dealing with very often so how do we manage like we have like one application definition we have to deploy to multiple stages which have different configurations and how do we handle an update how do we separate these kind of things like I think having it's almost like a challenge we create like an application a very simple one it doesn't need to be complex just need to have like these things in there and that's really how do you how do you like handle this challenge how handle this one I think that might be might be something interesting obviously we have to write something I think the code is not going to be super challenging this can just be a hello world that asks another hello world for its text so maybe it's two times ten lines of go code that's everything I say well but we want in this case the back end to be scaled 10 times and five times and here we wanted to be blue green deployed how do we do even specify this and here we want to do this and then we can work on scenarios and people can experiment and show their solutions and I think that could be then really something that is value to the cloud native community at harry that's actually where we're going kind of like with the that the talk was submitted for kubecon or the north america like really using projects to do different things and asking what people are dealing with but I think we have to bootstrap it I think we have to really bootstrap it and come up with two questions and ideally find people who we know who would be interested in doing it to get like this initial momentum because we try to have this let's reach out to the community and ask them to do something that was always a bit challenging when nothing is there but what we could do we could like tackle one problem and then like build the sample or maybe we just write up but samples are great because people like I would like to run a sample and sometimes it's easier to just see the actual code and then read a very long white paper we can then easily publish about it we can use cncf distributions to help to reach more audiences and I guess people see ah they're solving the use case that I'm having and also I think you will get deferral people from various open source and commercial companies to also contribute because usually that's what they do they say hey why are they showing this way of secrets management when we for example want to when when hashikov wants to see how vault can be used there and suddenly they have an example on how to be able to do it in vault and you can ideally build up a a library or that would expect that like we've had an opinion on how how github should be done so I think they would uh I hope they would find the resources to then then contribute they wouldn't make it like super complex and just putting it out there whether you think it makes sense to try something like this it's I love it I love it sample applications code code always speaks volumes that's a great idea and then the meetings could also be interactive somebody can do a demo and say yeah I have to demo it also for the future project okay this is the problem that you're solving show it based on the sample application not the sample application that you have but like eventually our reference application just show us what you can do there yeah but now we have just to put it out to to volunteers to get the first one running I maybe have sock shop and hipster shop so as I mentioned my only thing is there are two complex in the amount of services that they're running if you want to do especially multi-stage a bit resource hungry honestly and they don't cover all the aspects yeah that's a really good idea I can contribute some resources regarding to the demos and samples because we internally actually did a lot of demos I can actually try to ask those folks to contribute those back to the community and if you want to have some you know first round of the demos we are very happy to contribute that would be amazing I think and then we really have to actively share it out I think this if we have the demos I think we should cut the videos I mean not as professional usually as you do with Harry Harry's actually a magician with videos you can trust me working with him let's keep that to secret I'm just saying you're great you like the James Bond of making videos and yeah then also like share it on Twitter make it easily more easily available I think that's definitely something we should be doing and trying to get more reach that that would be great let's start with the demos the cleanup work is great as well and maybe then next time come to the conclusion what we do is just operate the dock maybe just leave it leave it as it is today maybe don't find any conclusion it's going to be still a CRD and a controller that's as good as you get sorry all right so Harry for a minute we ended up talking about a bunch of really good stuff now we just have to get it done that's always the challenging part here that's right we just have and we have to put it into the notes what I'm doing that's what I think finishing up the notes is what we should use the last 10 minutes for and I think Harry the two of us have no beauty and the rest of you will give you 10 minutes back of your time okay is that it for today then that's it yeah that that's it for today we're just finishing up the notes writing here that's what I meant before so you don't have to watch us typing so enough then thank you thank you very much I really appreciate it